Hi, 

I'm still working step by step on improving the digital version. 

Digital version in JavaScript - option to play against AI (heuristic and minimax), PvP game, solo mode with several difficulty levels.

My older son pointed out the solo mode to me, complaining that he often fails to complete the game board. We discussed the rules of solo mode and agreed that the game board cannot be randomly generated. Especially for solo mode, I redesigned the board generation using a solvable game board generator. The game log now includes a generated list of solutions, including a link to play back an animation of the generated solution for a given board. Playback can be canceled by pressing the Reset button or selecting another option from the list. The corrected list of moves, including intermediate steps, is now included in the game log.

Click on the game banner to display help.

Collapsi JS

I’m afraid you are misunderstanding the rules (which is my fault). Every board is solvable. I will have a better video soon. I know I did not do a good job explaining. 

You do not need an Ace beside the exit joker.  Think of the exit joker as a door. All you need to be able to do is pass through the door, so any number will work. 

Have you seen the most up-to-date rules for how to start the game?  Players must move from the joker to an adjacent card as their first move.  This includes cards they can reach through wrapping if their starting joker is on the edge of the board.  

Yes, the red joker leads the first move, and here the player can determine how many steps to use (1, 2, 3, 4), but then they must use as many steps as the value of the card they are standing on. That should correspond to your rules. But solo mode is something else. I originally played solo with a 4x4 board. After about 5 games, I managed to fold the board down to the black joker. But then I didn't manage it for a long time, and then I managed it again. But in my opinion, it was probably just a coincidence. So I modified the program to 3x3, and it's better here, but even so, there are combinations of 3x3 card layouts where the board is unplayable. That's why my son gave me the idea to generate game boards recursively. This means that the generation starts by placing the black joker, and then a card of any value is added to it, provided that it must be adjacent to it directly or via wrap-around. The next cards are added to the last card placed according to the same principle (they must be adjacent) and with a value that complies with the rules of the number of steps, wrap-around, and other rules. Finally, a red joker is placed on the last empty space on the game board, and the game board is recursively composed and, most importantly, always has a solution.

I don't quite understand the question. Are you considering a rule that in a solo game for one player, the red joker will also only be able to move one space, as in multiplayer?

I can't judge this, it completely changes the conditions of the game and it's more up to the AI and the creation of some kind of generator that would test the set conditions of the rules against the generated game board. 

Under the current conditions, I have two simple sets of cards (easy and medium difficulty) and five sets for hard difficulty in the program. However, there is a certain limitation here as well, in that not just any set of cards can be used.

I would rather summarize the rules precisely. Mainly, the game board for 2 or more players can be generated completely randomly, because the primary goal is to see who will be the first to have no legal move. Whereas solo mode is like a classic puzzle, it must always be solvable. And that's probably not possible with some combinations. I asked the AI and it gave me this example:

R 3 3

2 2 4

4 1 B

The restriction in solo mode to 1 move on the first turn, as intended in multiplayer, can have unexpected consequences for players because they will not be able to choose the card that seems most suitable to them on a freshly generated game board, but will be limited to the adjacent card (for the first move, this is a restriction to 3 moves - if adjacent to a black joker - or 4 moves instead of 7). It is really up to the AI to create a generator that evaluates this and determines whether the game board is solvable under these conditions or not. But to apply this rule in real life, when we shuffle 9 or, in the worst case, 16 cards and arrange them randomly, we have no idea whether it is playable under the given conditions.

I have been working with a solving program for a few months (with the rule that from the joker you can go to any card in the grid).  Trust me, these puzzles can be solved as long as there is a 1 that can access the exit joker. =^)

The solver has not been changed to account for the potential rule change that the player must move to an adjacent card from the starting joker.

In regard to your app, when I play the multiplayer game, it is not beginning correctly.  From the starting joker, I can move to any card on the board.  The updated rules state that you must move from the joker to an adjacent card.  The starting joker is essentially a 1 now.  I think you said the app plays that way, but it's not doing that for me.  When playing under all of the multiplayer modes, the piece can be moved to any card on the board from the starting joker.

Hi, thanks for the feedback. I misunderstood when I saw you talking about 3- and 4-player games in the chat.

I modified the program according to the rules so that the first move of both players is only 1 step, and I fixed the situation so that a joker cannot move onto another joker. The second and subsequent moves are governed by the rule of the value of the card I land on, and I have the corresponding number of steps available in my move.

I'm leaving the solo game in the program because I really like it, and the generator I use creates solvable board variations. See the game logo image  (Generated solution path). Try click to replay solution.

Collapsi JS

Added button to switch AI mode settings minimax:

0 - minimax

2 - first 2 moves heuristics → minimax

3 - first 3 moves heuristics → minimax